app.py

import numpy as np
from PIL import Image
import gradio as gr
from deepface import DeepFace
from datasets import load_dataset
import os
import pickle
from pathlib import Path
import gc
import requests
from io import BytesIO

# 📁 Directorio local para embeddings
EMBEDDINGS_DIR = Path("embeddings")
EMBEDDINGS_DIR.mkdir(exist_ok=True)
EMBEDDINGS_FILE = EMBEDDINGS_DIR / "embeddings.pkl"

headers = {}
HF_TOKEN = os.getenv("HF_TOKEN")
if HF_TOKEN:
    headers["Authorization"] = f"Bearer {HF_TOKEN}"



# ✅ Cargar el dataset remoto desde Hugging Face Datasets con metadata.csv
dataset = load_dataset(
    "csv",
    data_files="metadata.csv",
    split="train",
    column_names=["image"],
    header=0  # 👈 asegúrate de que la primera fila se trate como encabezado
)

print("✅ Validación post-carga")
print(dataset[0])
print("Columnas:", dataset.column_names)

print("✅ Primeros ítems de validación:")
for i in range(5):
    print(dataset[i])

# 🔄 Preprocesar imagen para DeepFace
def preprocess_image(img: Image.Image) -> np.ndarray:
    img_rgb = img.convert("RGB")
    img_resized = img_rgb.resize((160, 160), Image.Resampling.LANCZOS)
    return np.array(img_resized)

# 📦 Construir base de datos de embeddings
def build_database():
    if EMBEDDINGS_FILE.exists():
        print("📂 Cargando embeddings desde archivo...")
        with open(EMBEDDINGS_FILE, "rb") as f:
            return pickle.load(f)

    print("🔄 Calculando embeddings...")
    database = []
    batch_size = 10

    for i in range(0, len(dataset), batch_size):
        batch = dataset[i:i + batch_size]
        print(f"📦 Procesando lote {i // batch_size + 1}/{(len(dataset) + batch_size - 1) // batch_size}")

        for j in range(len(batch["image"])):
            try:
                item = {"image": batch["image"][j]}

                image_url = item["image"]
                if not isinstance(image_url, str) or not image_url.startswith("http") or image_url.strip().lower() == "image":
                    print(f"⚠️ Saltando item {i + j} - URL inválida: {image_url}")
                    continue

                # Autenticación para datasets privados
                headers = {}
                HF_TOKEN = os.getenv("HF_TOKEN")
                if HF_TOKEN:
                    headers["Authorization"] = f"Bearer {HF_TOKEN}"

                response = requests.get(image_url, headers=headers, timeout=10)
                response.raise_for_status()
                img = Image.open(BytesIO(response.content)).convert("RGB")

                img_processed = preprocess_image(img)
                embedding = DeepFace.represent(
                    img_path=img_processed,
                    model_name="Facenet",
                    enforce_detection=False
                )[0]["embedding"]

                database.append((f"image_{i + j}", img, embedding))
                print(f"✅ Procesada imagen {i + j + 1}/{len(dataset)}")

                del img_processed
                gc.collect()

            except Exception as e:
                print(f"❌ Error al procesar imagen {i + j}: {str(e)}")
                continue

    # Guardar al final si hay datos
    if database:
        print("💾 Guardando embeddings finales...")
        with open(EMBEDDINGS_FILE, "wb") as f:
            pickle.dump(database, f)

    return database

# 🔍 Buscar rostros similares
def find_similar_faces(uploaded_image: Image.Image):
    try:
        img_processed = preprocess_image(uploaded_image)
        query_embedding = DeepFace.represent(
            img_path=img_processed,
            model_name="Facenet",
            enforce_detection=False
        )[0]["embedding"]
        del img_processed
        gc.collect()
    except Exception as e:
        print(f"Error al procesar imagen de entrada: {str(e)}")
        return [], "⚠ No se detectó un rostro válido."

    similarities = []
    for name, db_img, embedding in database:
        dist = np.linalg.norm(np.array(query_embedding) - np.array(embedding))
        sim_score = 1 / (1 + dist)
        similarities.append((sim_score, name, db_img))

    similarities.sort(reverse=True)
    top_matches = similarities[:5]

    gallery_items = []
    summary = ""
    for sim, name, img in top_matches:
        caption = f"{name} - Similitud: {sim:.2f}"
        gallery_items.append((np.array(img), caption))
        summary += caption + "\n"

    return gallery_items, summary

# 🚀 Iniciar aplicación
print("🚀 Iniciando aplicación...")
database = build_database()
print(f"✅ Base cargada con {len(database)} imágenes.")

# 🎛️ Gradio UI
demo = gr.Interface(
    fn=find_similar_faces,
    inputs=gr.Image(label="📤 Sube una imagen", type="pil"),
    outputs=[
        gr.Gallery(label="📸 Rostros más similares"),
        gr.Textbox(label="🧠 Similitud", lines=6)
    ],
    title="🔍 Buscador de Rostros con DeepFace",
    description="Sube una imagen y se comparará contra los rostros del dataset `Segizu/facial-recognition` almacenado en Hugging Face Datasets."
)

demo.launch()